YouTube should show you more of what you like. If it doesn’t, you can nudge it back to relevancy.
If the site is showing you irrelevant suggestions in the right sidebar, at the end of videos and on the homepage, then you’re likely giving it the wrong idea about what you like — or not providing enough signals at all. It might take some time to retrain, but you can teach YouTube to show more of what you enjoy and less of what you don’t.
First, make sure watch history is enabled
To enable this setting, click your profile icon at the top-right of YouTube and choose Google Account. On the Google Account page, select Data & privacy from the left sidebar, then choose Personalization settings. In this menu, find YouTube History and make sure it’s enabled. The Subsettings checkboxes are optional but provide more data points to use in suggestions.
From this box, check the status of Auto-delete too. Google gives you the option to automatically clear this data every three, 18 or 36 months. Doing so means less information to use for recommendations, but it also gives the algorithm a more focused data set to work with.
The key step: Show YouTube what you like and dislike
Favorable input is important, but you can provide YouTube with negative feedback too. Hit Dislike on videos that are uninteresting or waste your time. When you see a recommendation you don’t want to watch, click the three-dot (…) button on it and choose Not interested. If it’s the entire channel you don’t like, choose Don’t recommend channel instead. Be sure to unsubscribe from channels you don’t watch anymore. And remember that hiding YouTube Shorts is its own option.
The negative signals aren’t perfect; you might see the same recommendations again after a while, and it takes time for the algorithm to properly shift. But combining these two will help teach the service what you’re likely to click on.
Minimize incorrect watching
However, it’s better to prevent this from happening in the first place. If your family watches YouTube using the TV app often, for instance, make sure you have everyone’s accounts added so nobody’s watching habits affect anyone else’s suggestions. There’s also an option to use the app without signing in, if needed. In case your kids keep picking the wrong account, you can lock yours (separate from YouTube’s parental controls). In the YouTube app for TVs, open the account page by selecting your icon from the left bar, then choose the Settings gear underneath your account. Use the Account lock option to set a code that you must enter to use that account.
When you’re about to watch videos that you don’t want to influence your recommendations on a desktop or laptop, use private browsing. In the mobile app, go to You > Accounts > Turn on Incognito. A stray video here or there shouldn’t have a huge effect, but this can proactively prevent you from sending the wrong signals to YouTube.
Manage your watch history (selectively)
This tool also holds a Delete button to mass-remove your historical data from the service. You have options to delete data from today, a specific time period or everything. The specific time periods are suitable when you have a block of unwanted watching, like if you’ve been troubleshooting a car problem but don’t otherwise care about auto videos.
You might have seen advice to completely wipe your YouTube history to improve recommendations, but whether you should do this depends on your situation. If your suggestions are completely terrible and nothing you see is at all relevant, then nuking and starting over can work. When you start from nothing, YouTube shows you the same “broad appeal” videos you see when you aren’t signed in, and it takes time to build up a revamped profile. If you go this route, it’s smart to have a list of channels you like on hand so you can start retraining YouTube right away after clearing.
Otherwise, more selective deletion offers a better balance. If you were really into gardening six months ago but have fallen out of love with it recently, use the custom range option to clear history from that period while keeping more recent data. From there, giving YouTube those discussed positive and negative signals should result in fewer poor recommendations, and hopefully fewer AI slop videos, over time.

